CN110134552B - Fault-tolerant method based on empirical learning - Google Patents
Fault-tolerant method based on empirical learning Download PDFInfo
- Publication number
- CN110134552B CN110134552B CN201910386756.7A CN201910386756A CN110134552B CN 110134552 B CN110134552 B CN 110134552B CN 201910386756 A CN201910386756 A CN 201910386756A CN 110134552 B CN110134552 B CN 110134552B
- Authority
- CN
- China
- Prior art keywords
- channels
- voting
- fault
- signals
- results
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
- G06F11/16—Error detection or correction of the data by redundancy in hardware
- G06F11/1629—Error detection by comparing the output of redundant processing systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
- G06F11/16—Error detection or correction of the data by redundancy in hardware
- G06F11/1629—Error detection by comparing the output of redundant processing systems
- G06F11/1654—Error detection by comparing the output of redundant processing systems where the output of only one of the redundant processing components can drive the attached hardware, e.g. memory or I/O
Abstract
The application provides a fault-tolerant method based on empirical learning, which is used for acquiring signals of N channels; voting the signals of the N channels respectively to obtain voting results of the N channels; and voting the signals of the N channels according to the voting results of the N channels and the credibility evaluation of the N channels.
Description
Technical Field
The invention belongs to the field of design of a strong real-time and high-reliability control system, and relates to a fault-tolerant strategy of a four-redundancy fault-tolerant computer.
Background
In order to ensure the reliability of a strong real-time and high-reliability control system, a redundancy fault-tolerant design method can be adopted, for example, in the control of high lift control, flight control and an engine of an aircraft, the reliability of the control system can be improved by adopting fault-tolerant designs with different redundancies. The key technology of fault tolerant computers is fault tolerance strategy. The goal of the fault tolerance strategy is to determine faults in the system and reconfigure the system, isolate the faults, and prevent the propagation of the faults.
In a quad-redundancy fault-tolerant computer, the traditional majority voting fault-tolerant strategy is as follows, 2:2, uncertainty is generated under the voting result, and the uncertainty destroys the strong real-time property of a strong real-time and high-reliability control system and influences the timeliness of a critical task.
Disclosure of Invention
Aiming at the characteristics of a strong real-time and high-reliability control system, a fault-tolerant strategy based on empirical learning of a four-redundancy fault-tolerant computer is provided.
In a first aspect, the application provides a fault-tolerant method based on empirical learning, which acquires signals of N channels;
voting the signals of the N channels respectively to obtain voting results of the N channels;
and voting the signals of the N channels according to the voting results of the N channels and the credibility evaluation of the N channels.
Optionally, the voting result of the N channels includes:
voting results obtained by voting input signals, voting results obtained by processing input signals, and voting results obtained by voting output signals.
Optionally, if N is an even number, voting the signals of the N channels according to the voting results of the N channels and the reliability evaluations of the N channels includes:
when N/2: and when the voting results with the same number of votes of N/2 are obtained, adopting the voting results with high reliability.
Optionally, if N is an odd number, voting the signals of the N channels according to the voting results of the N channels and the reliability evaluations of the N channels, specifically including:
and adopting a large number of voting results.
Optionally, if the votes of the different results of the N channels are the same, voting the signals of the N channels according to the voted results of the N channels and the reliability evaluations of the N channels, specifically including:
and adopting the voting result with high reliability.
Optionally, the reliability evaluation of the N channels specifically includes:
recording the failure frequency of the N channels;
and respectively carrying out credibility evaluation based on experience learning on the N signal channels according to the fault frequency.
Optionally, the fault-tolerant method is applied to a redundancy fault-tolerant computer platform.
Optionally, N is 4.
In summary, the present invention can eliminate the occurrence of the four-redundancy fault-tolerant architecture 2:2 uncertainty in voting the result. Through experience learning, the credibility evaluation of the input stage, the processing stage and the output stage of each channel is realized, and when 2: and 2, when voting the result, evaluating and adopting data with high reliability according to the reliability of empirical learning. The reliability evaluation based on experience learning is provided, fault isolation and non-fault isolation (isolation of data with low reliability) are realized, and more reliable data extraction is facilitated.
Description of the drawings:
FIG. 1 is a flowchart of a confidence evaluation provided by an embodiment of the invention;
fig. 2 is a block diagram of a system architecture according to an embodiment of the present invention.
Detailed Description
Aiming at the characteristics of a strong real-time and high-reliability control system and a four-redundancy fault-tolerant computer, a fault-tolerant strategy based on experience learning under the four-redundancy fault-tolerant computer is provided, and fault identification, fault isolation and low-reliability isolation of software and hardware faults of the strong real-time and high-reliability control system and the four-redundancy fault-tolerant computer are realized.
Under a four-redundancy fault-tolerant architecture, a traditional majority voting fault-tolerant strategy appears in 2:2 voting results can create uncertainties that undermine the strong real-time nature of the control system. Based on the fault-tolerant strategy of empirical learning, the fault-tolerant strategy can select the channel with high reliability by recording the fault frequency of each channel, and the result is adopted. Thanks to the experience learning, the reliability of the four-redundancy fault-tolerant computer is improved along with the increase of the service time.
Example one
As shown in fig. 1, a fault-tolerant strategy based on empirical learning is implemented based on a four-redundancy fault-tolerant computer, and according to the empirical learning, the frequency of faults of each channel is recorded, so as to evaluate the credibility of each channel, and the fault-tolerant strategy includes the following steps:
collecting signals such as discrete quantity, analog quantity and digital quantity by 4 channels in a four-redundancy fault-tolerant computer, majority voting input signals, adding 1 to the record count of the channel corresponding to the collected fault in a reliability list, and performing reliability evaluation based on empirical learning on each signal channel after multiple times of collection and voting;
processing units of 4 channels in a four-redundancy fault-tolerant computer process input signals, processing results are transmitted in a cross mode through internal communication channels, majority voting is carried out on the processing results, the recording count of a corresponding channel of an error processing unit in a credibility list is increased by 1, and credibility evaluation based on experience learning is carried out on each processing unit through multiple times of calculation and voting;
carrying out majority voting on the results of output signals of 4 channels in the quad-redundancy fault-tolerant computer, adding 1 to the record count of the channel corresponding to the output fault in a reliability list, and carrying out reliability evaluation based on experience learning on each output channel after multiple times of output and voting;
in the input stage, processing stage and output stage of a four-redundancy fault-tolerant computer, when 2:2, adopting the voting result with high reliability when voting the result;
the application platform and application opportunities of the fault-tolerant strategy based on empirical learning are limited.
Preferably, the fault-tolerant strategy based on empirical learning is limited in application platform and application time, and is characterized in that the fault-tolerant strategy is applied to a four-redundancy similar/dissimilar fault-tolerant computer platform, and the application time is as follows: the computer platform accumulates 2: and 2, judging the credible data under the voting condition.
Preferably, the reliability evaluation is characterized in that fault channels of all levels (input level, processing level and output level) are recorded according to a majority voting result of the quad-redundancy fault-tolerant computer, and the reliability of the channel with the higher count in the reliability list is lower, so that the reliability evaluation based on empirical learning is formed.
Preferably, the input signal and the output signal are characterized in that each channel of the four-redundancy fault-tolerant computer can input the signal and each channel can output the signal.
Example two
The present invention is described in further detail below.
As shown in fig. 1, a typical composition of a quad-redundant fault tolerant computer includes: 4 channels with the same input function, processing function and output function, and the software and hardware of the 4 channels can be similar or dissimilar. The traditional majority voting fault-tolerant strategy can realize fault identification and fault isolation in most cases, but under a four-redundancy framework, 2: the uncertainty generated by the voting result destroys the real-time property of a strong real-time and high-reliability system, and the uncertainty can be eliminated by adopting a fault-tolerant strategy based on empirical learning according to the credibility evaluation of each channel at each stage (input stage, processing stage and output stage). The core part of the fault tolerance strategy based on empirical learning is credibility evaluation. Credibility assessment is formed based on empirical learning, and the specific flow thereof is shown in fig. 1.
In the majority voting comparison, regarding the analog quantity, the analog quantity meeting a certain tolerance range is considered to be equal, and the discrete quantity and the digital quantity are compared in a consistent manner. When majority voting occurs at each stage (input stage, processing stage, and output stage), when 3:1, adding 1 to the record count of the functional stages (input stage, processing stage and output stage) corresponding to the fault and the corresponding channels in the credibility list. After each function level of each channel is subjected to experience learning for multiple times, the failure frequency of each function level of each channel is recorded in the reliability list, and a reliability evaluation table based on the experience learning is formed. When 2: when voting results are 2 (e.g., a = B ≠ C = D), the failure frequency of the corresponding functional level of the channel in the reliability list is read and compared (e.g., μ 1= a + B > μ 2= C + D), and the reliability of the pair (a, B) is lower than that of the pair (C, D), so the result of the pair (C, D) is adopted as a correct result.
The fault tolerance strategy based on empirical learning only records 3:1, voting faults under the result, and improving the credibility of the credibility list; since the confidence level list is formed by empirical learning, the longer the time of empirical learning for a four-redundancy fault-tolerant computer, the higher the confidence level of the confidence level list.
The system implementation principle is shown in fig. 2.
Claims (4)
1. A fault-tolerant method based on empirical learning is characterized in that,
acquiring signals of N channels;
voting the signals of the N channels respectively to obtain voting results of the N channels;
voting the signals of the N channels according to the voting results of the N channels and the credibility evaluation of the N channels;
the voting results of the N channels comprise: voting results obtained by voting the input signals, voting results obtained by processing the input signals, and voting results obtained by voting the output signals;
if N is an even number, voting the signals of the N channels according to the voting results of the N channels and the reliability evaluations of the N channels, which specifically includes: when N/2: adopting the voting result with high reliability when the voting results with the same number of votes of N/2 are obtained;
if N is an odd number, voting the signals of the N channels according to the voting results of the N channels and the reliability evaluations of the N channels, which specifically includes: adopting a large number of voting results;
the credibility evaluation of the N channels specifically includes:
recording the failure frequency of the N channels;
and respectively carrying out credibility evaluation based on experience learning on the N signal channels according to the fault frequency.
2. The method of claim 1, wherein if the votes for the different results of the N channels are the same, voting the signals of the N channels according to the voted results of the N channels and the confidence evaluations of the N channels comprises:
and adopting the voting result with high reliability.
3. The method of claim 1,
the fault-tolerant method is applied to a redundancy fault-tolerant computer platform.
4. The method of claim 1, wherein N is 4.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910386756.7A CN110134552B (en) | 2019-05-09 | 2019-05-09 | Fault-tolerant method based on empirical learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910386756.7A CN110134552B (en) | 2019-05-09 | 2019-05-09 | Fault-tolerant method based on empirical learning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110134552A CN110134552A (en) | 2019-08-16 |
CN110134552B true CN110134552B (en) | 2023-04-14 |
Family
ID=67577041
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910386756.7A Active CN110134552B (en) | 2019-05-09 | 2019-05-09 | Fault-tolerant method based on empirical learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110134552B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110687775B (en) * | 2019-10-11 | 2022-06-07 | 中国航发沈阳发动机研究所 | Dual-redundancy sensor signal voting method based on fault integration |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1214828A (en) * | 1996-04-01 | 1999-04-21 | 艾利森公司 | Method and apparatus for data recovery in ARQ systems |
JP2006146293A (en) * | 2004-11-16 | 2006-06-08 | Dainippon Printing Co Ltd | Data storage mediating system |
CN103870353A (en) * | 2014-03-18 | 2014-06-18 | 北京控制工程研究所 | Multicore-oriented reconfigurable fault tolerance system and multicore-oriented reconfigurable fault tolerance method |
CN105182961A (en) * | 2015-08-11 | 2015-12-23 | 中国航空工业集团公司西安飞机设计研究所 | Quad-redundancy signal monitoring and voting method and quad-redundancy signal monitoring and voting equipment |
CN105487945A (en) * | 2016-02-19 | 2016-04-13 | 中国航天科技集团公司第五研究院第五一三研究所 | Self-monitoring fault-tolerant control method of non-similar dual-redundancy four processors |
CN105629797A (en) * | 2016-03-29 | 2016-06-01 | 杭州和利时自动化有限公司 | Output control method and system for N output channels |
CN108415238A (en) * | 2018-02-02 | 2018-08-17 | 中国航空工业集团公司沈阳飞机设计研究所 | A kind of electricity trim monitoring switch method |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100169886A1 (en) * | 2008-12-31 | 2010-07-01 | Seakr Engineering, Incorporated | Distributed memory synchronized processing architecture |
-
2019
- 2019-05-09 CN CN201910386756.7A patent/CN110134552B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1214828A (en) * | 1996-04-01 | 1999-04-21 | 艾利森公司 | Method and apparatus for data recovery in ARQ systems |
JP2006146293A (en) * | 2004-11-16 | 2006-06-08 | Dainippon Printing Co Ltd | Data storage mediating system |
CN103870353A (en) * | 2014-03-18 | 2014-06-18 | 北京控制工程研究所 | Multicore-oriented reconfigurable fault tolerance system and multicore-oriented reconfigurable fault tolerance method |
CN105182961A (en) * | 2015-08-11 | 2015-12-23 | 中国航空工业集团公司西安飞机设计研究所 | Quad-redundancy signal monitoring and voting method and quad-redundancy signal monitoring and voting equipment |
CN105487945A (en) * | 2016-02-19 | 2016-04-13 | 中国航天科技集团公司第五研究院第五一三研究所 | Self-monitoring fault-tolerant control method of non-similar dual-redundancy four processors |
CN105629797A (en) * | 2016-03-29 | 2016-06-01 | 杭州和利时自动化有限公司 | Output control method and system for N output channels |
CN108415238A (en) * | 2018-02-02 | 2018-08-17 | 中国航空工业集团公司沈阳飞机设计研究所 | A kind of electricity trim monitoring switch method |
Non-Patent Citations (1)
Title |
---|
四余度飞控计算机信号表决与监控算法研究;张锐;《信息系统工程》;20180630(第6期);第142-144页 * |
Also Published As
Publication number | Publication date |
---|---|
CN110134552A (en) | 2019-08-16 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2019184557A1 (en) | Method and device for locating root cause alarm, and computer-readable storage medium | |
Southern et al. | Deriving ICD-10 codes for patient safety indicators for large-scale surveillance using administrative hospital data | |
AU660661B2 (en) | Knowledge based machine initiated maintenance system | |
Godbout et al. | Legislative voting in the Canadian Parliament | |
US8751417B2 (en) | Trouble pattern creating program and trouble pattern creating apparatus | |
Salfner et al. | Error Log Processing for Accurate Failure Prediction. | |
Hong et al. | When is acceleration unnecessary in a degradation test? | |
CN110134552B (en) | Fault-tolerant method based on empirical learning | |
US10169112B2 (en) | Event sequence management | |
US11429574B2 (en) | Computer system diagnostic log chain | |
CN110708339B (en) | Correlation analysis method based on WEB log | |
US8996825B2 (en) | Judgment apparatus, judgment method, and recording medium of judgment program | |
CN112951311A (en) | Hard disk fault prediction method and system based on variable weight random forest | |
Kaashoek et al. | The evolving roles of US political partisanship and social vulnerability in the COVID-19 pandemic from February 2020–February 2021 | |
CN105740361A (en) | Detection method and apparatus for integrity of total data | |
CN116756518A (en) | Cloud edge cooperative power defect identification method and related device based on federal learning | |
CN116663136A (en) | Method for evaluating reliability confidence lower limit of ship propulsion system series structure | |
Zahid et al. | Trends, predictors, and outcomes of cardiovascular complications at delivery associated with gestational diabetes: a National Inpatient Sample analysis (2004–2019) | |
CN115700549A (en) | Model training method, failure determination method, electronic device, and program product | |
Ansolabehere et al. | Using Recounts to Measure the Accuracy of Vote Tabulations: Evidence from New Hampshire Elections 1946-2002 | |
Pu et al. | Challenges and opportunities in rapid epidemic information propagation with live knowledge aggregation from social media | |
AT&T | main.dvi | |
CN109426576A (en) | Fault-tolerance processing method and fault-tolerant component | |
Knotts et al. | The nationalization of special elections for the US House of representatives | |
KR101763394B1 (en) | Apparatus for monitoring streaming data processing system and method thereof |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |